Objectives To compare the prevalence of medial and lateral patellofemoral (PF) cartilage damage in three large osteoarthritis (OA) studies and determine the relationship of this damage to varus, neutral and valgus knee alignment.
Methods In the Boston OA of the Knee, Framingham OA and Multicenter OA studies, MRIs were read for cartilage morphology at the medial and lateral patella and trochlea femoris using Whole-Organ MRI Scores (WORMS). WORMS scores ≥2 (any cartilage defect), ≥3 (areas of partial thickness loss), ≥4 (diffuse partial thickness loss) and ≥5 (extensive full thickness loss) were all variously considered as thresholds to identify damage that may indicate OA. Full-limb radiographs were measured for mechanical alignment, and varus (<−2°), neutral (-2° to 2°) and valgus (>2°) knees were identified.
Results The prevalence of medial PF cartilage damage exceeded that of lateral damage in all three studies and according to nearly every threshold. Only among severely involved knees (WORMS ≥4 or ≥5) did the prevalence of lateral PF cartilage damage approximate that of medial damage. The high prevalence of medial PF damage persisted in all strata of knee alignment. Even among knees with valgus alignment, the prevalence of lateral PF cartilage damage equalled or surpassed that of medial PF damage only when the threshold was specific to severely involved knees.
Conclusions Medial PF cartilage damage is at least as prevalent within these older adult populations as lateral PF cartilage damage.
Statistics from Altmetric.com
In their classic textbook, Ficat and Hungerford introduced the ‘Law of Valgus.’1 With reference to the valgus quadriceps angle (Q-angle), the Law of Valgus dictates that the predominant frontal plane force on the patella is directed laterally (figure 1). It followed that patellofemoral (PF) osteoarthritis (OA) was principally a disease of excessive loading and cartilage damage in the lateral PF compartment. Since Ficat and Hungerford, others2,–,5 have repeated the assertion that the only circumstances under which medial PF OA should arise with any frequency are those in which varus knee malalignment has supplanted the typical valgus Q-angle with an atypical varus Q-angle.
Despite widespread acceptance of the Law of Valgus, only one small Japanese study6 has corroborated these predictions. In a study of 44 knees with PF OA on skyline radiograph, only seven had medial disease. Among these seven, all but one had varus knee malalignment. Thus, the findings of this small x-ray study seemed to support the assertion that medial PF OA is an uncommon condition which, when present, is almost invariably a consequence of varus knee malalignment.
Our clinical experience contradicts these conclusions. We have found that many patients report medial peripatellar pain and have medial retropatellar tenderness on examination. This experience is paralleled by reports from cadaver studies that fibrillation and other signs of cartilage damage are frequent at the medial patellar facet.7 ,8
With the advent of MRI and large scale OA studies, it is possible to compare the prevalence of medial and lateral PF cartilage damage directly. Such an evaluation could test assertions about the low prevalence of medial PF disease and inform biomechanical treatment strategies.
Our objective was to compare the prevalence of medial and lateral PF cartilage damage using MRI measures of cartilage morphology in three large OA studies. We sought to better understand the prevalence of medial and lateral PF damage by determining their relationship to varus, neutral and valgus alignment.
We determined the prevalence of medial and lateral PF cartilage damage, and of isolated medial, isolated lateral, and mixed medial and lateral PF cartilage damage within the Boston Osteoarthritis of the Knee Study (BOKS), Framingham Osteoarthritis (FOA) study and Multicenter Osteoarthritis Study (MOST). In each study, MRIs were read for cartilage morphology at the medial and lateral patella, and medial and lateral trochlea femoris using semiquantitative Whole-Organ MRI Scores (WORMS).9 Bilateral standing full-limb anteroposterior radiographs were acquired with the knee fully extended. Mechanical axis alignment was measured to the nearest 0.1° using an established protocol,10 and varus (<−2°), neutral (-2° to 2°) and valgus (>2°) aligned knees were identified.
In BOKS, radiographic PF OA (PF ROA) was identified on skyline x-ray. In FOA and MOST, lateral knee x-rays were acquired in a semiflexed weight-bearing position and read using validated methods.11 Radiographic tibiofemoral OA (TF ROA) was present when Kellgren and Lawrence grade was ≥2 on semiflexed anteroposterior knee x-ray.
Boston Osteoarthritis of the Knee Study
BOKS was a natural history study of symptomatic knee OA based at the Veterans Affairs (VA) Medical Center in Boston, Massachusetts, USA, and approved by Institutional Review Boards (IRBs) at the VA and Boston University Medical Center (BUMC). Participants had at least one knee that met the American College of Rheumatology (ACR) criteria for knee OA. Patients with inflammatory arthritis were excluded.12
MRIs of the more symptomatic knee were acquired at the baseline visit using a Signa 1.5 Tesla (T) magnet (General Electric Corp., Milwaukee, Wisconsin, USA) and phased-array knee coil. A positioning device ensured consistent positioning. The protocol included axial, sagittal and coronal spin-echo fat-suppressed proton-density T2-weighted images (repetition time (TR) 2200 ms; time to echo (TE) 20/80 ms; slice thickness (ST) 3 mm; 0.3 mm interslice gap (IG); 256×192 matrix; 120 mm2 field of view (FOV); echo train length (ETL) 1). Full-limb radiographs were obtained at the first BOKS follow-up visit, and read for knee alignment by two readers (inter-reader Intraclass Correlation Coefficient (ICC) = 0.97)
Framingham Osteoarthritis study
Participants in FOA were from two subgroups. The Offspring cohort13 included offspring (and their spouses) of participants in the original Framingham Heart Study.14 The Community cohort15,–,17 was a recent population-based sample of the Framingham community.18 Participants were 50–80 years old and ambulatory (canes and walkers permitted). As in BOKS, a validated survey excluded patients with inflammatory arthritis.12 Unlike BOKS, neither FOA subgroup was selected for the presence of knee pain or OA.15 Approval was obtained from the BUMC IRB.
MRIs were acquired using a 1.5 T magnet (Siemens Symphony, Erlangen, Germany), and the right knee was read. If the right knee had a prosthesis, the left knee was imaged and read. In the Offspring cohort, only participants reporting knee symptoms underwent MRI. In contrast, MRIs were acquired in all Community participants, regardless of symptoms. The protocol included axial, sagittal and coronal turbo spin-echo fat-suppressed intermediate-weighted sequences (TR 3610 ms, TE 40 ms, 3.5 mm ST, 0 mm IG, 256×256 matrix, 139 mm2 FOV, ETL 6).16 ,17 A positioning device ensured consistent positioning.
Full-limb radiographs were obtained from all Community participants, but read for alignment only among a subsample previously identified for inclusion in a case-control study of knee OA.19 Where possible, we report results for both the entire FOA cohort and for this subsample. Within the subsample, a single reader (intrareader ICC=0.99) measured knee alignment.
Multicenter Osteoarthritis Study
MOST is a prospective study whose goal is to identify risk factors for incident and progressive knee OA in a large cohort of adults aged 50–79 from Birmingham, Alabama and Iowa City, Iowa, USA, who have or are at risk for knee OA. At-risk persons were overweight or obese, had frequent knee symptoms, or a history of knee injury or surgery.20 IRBs at the University of Iowa, University of Alabama, University of California, and BUMC approved the study.
MRIs of both knees were obtained from all participants at the baseline visit using a 1.0 T magnet (OrthOne, ONI Inc., Wilmington, Massachusetts, USA) and circumferential extremity coil. The protocol included axial, sagittal and coronal turbo spin-echo fat-suppressed intermediate-weighted sequences (TR 4800 ms, TE 35 ms, 3 mm ST, 0 mm IG, 288×192 matrix, 140 mm2 FOV, ETL 8). Readings were completed on a random sample of the progression subcohort and several case-control samples of the incident subcohort. Full-limb radiographs were obtained from all participants at the baseline visit. Two readers (inter-reader ICC=0.99) measured knee alignment.
MRI readings for PF cartilage damage
In BOKS, FOA and MOST, the same two radiologists (AG and Frank W Roemer ) executed identical reading protocols. To characterise PF cartilage damage and its location, they followed recommendations of the Workshop for Consensus on OA Imaging by focusing on the axial and sagittal views.21 Cartilage morphology was graded separately on the medial and lateral patellar and trochlear surfaces using WORMS.9 WORMS cartilage scores range from 0 to 6, as follows: 0=normal thickness and signal; 1=normal thickness but increased signal on T2-weighted images; 2.0=partial thickness focal defect <1 cm in greatest width; 2.5=full-thickness focal defect <1 cm in greatest width; 3=multiple areas of partial thickness (grade 2.0) defects intermixed with areas of normal thickness, or grade 2.0 defect >1 cm but <75% of the region; 4=diffuse (≥75% of the region) partial thickness loss; 5=multiple areas of full-thickness loss (grade 2.5) or a grade 2.5 lesion >1 cm but <75% of the region; and 6=diffuse (≥75% of the region) full-thickness loss. During blinded reassessment, weighted κ for inter-reader reliability in WORMS scoring of patellar and trochlear regions was 0.63 (95% CI 0.59 to 0.67) among 67 knees in BOKS, 0.73 (0.71 to 0.75) among 170 knees in FOA and 0.78 (0.73 to 0.82) among 10 knees in MOST.
WORMS demarcations assign the nadir of the patella (crista) to the medial patella, even though the crista may not belong to either the medial or lateral PF compartments. To assess the extent to which findings may have been influenced by this aspect of the protocol, we had one radiologist (AG), who was unaware of the hypothesis being tested, reread a random selection of 26 MRIs from among knees in FOA previously identified as having WORMS ≥2 medial patellar cartilage damage. During this rereading, we instructed him to score the medial patella separately from the crista.
There is currently no consensus specifying how severely cartilage must be damaged before OA is diagnosed on MRI. We applied four WORMS score thresholds to identify cartilage damage that might indicate OA, and compared the prevalence of medial and lateral PF disease using each of these cut-points. WORMS scores ≥2, ≥3, ≥4 and ≥5 were all variously considered.
Within each cohort and at each WORMS score threshold, a χ2 test was conducted to inform crude comparisons of: (1) medial to lateral PF cartilage damage prevalence, (2) isolated medial to isolated lateral PF cartilage damage prevalence and (3) medial to lateral PF cartilage damage prevalence within strata of varus, neutral and valgus aligned knees.
All three study populations had mean age in the mid-60s (table 1). Consistent with the fact that BOKS participants had at least one knee that met ACR criteria for OA, the BOKS study had a higher prevalence of PF ROA (64.0% vs. 15.7% in FOA and 18.0% in MOST), TF ROA (74.5% vs. 44.2% in FOA and 40.2% in MOST), and knee pain (94.0% vs. 29.6% in FOA and 33.2% in MOST). Within the FOA study, PF ROA, TF ROA and knee pain were more prevalent among those with alignment measures (table 1, asterisks). Because BOKS participants were mainly military veterans, the proportion of females was lower in BOKS than in other studies. Mean body mass index was high in all three studies.
The prevalence of medial PF cartilage damage exceeded that of lateral damage in all cohorts and according to nearly every threshold definition (table 2). When applying a threshold commonly used to identify cartilage damage suggestive of OA (WORMS ≥2), the prevalence of medial PF damage exceeded that of lateral damage (p<0.01) by 11%–19% across the cohorts. This was true regardless of whether the patella alone was considered (table 2, left), or both patella and trochlea were considered (table 2, right). Only when consideration was limited to the most severely involved knees (WORMS ≥4 or ≥5) did the prevalence of lateral PF cartilage damage approximate that of medial PF damage.
In comparisons of the prevalence of isolated medial, isolated lateral, and mixed medial and lateral PF cartilage damage (table 3), the most common finding was that of mixed damage. However, the prevalence of isolated medial damage still exceeded that of isolated lateral damage in most instances. Again, it was only when the most restrictive thresholds were applied (WORMS ≥4 or ≥5) that the prevalence of isolated lateral PF damage approximated that of isolated medial PF damage. In all studies, medial PF cartilage damage remained as or more prevalent than lateral damage within separate body mass index (normal or underweight vs overweight or obese), age (<70 vs >70 years) and gender strata.
We then determined the extent to which the high prevalence of medial PF cartilage damage could be explained by its association with varus knee malalignment (figure 2 and online supplementary table S1). Among knees with varus malalignment (<−2°), there was indeed a higher prevalence of medial than lateral PF cartilage damage in all cohorts and according to every threshold definition (figure 2, left). Yet, even among knees with neutral alignment (-2° to 2°), the high prevalence of medial damage persisted (figure 2, middle). More striking was the high prevalence of medial PF damage among knees with valgus malalignment (>2°) (figure 2, right). Contrary to expectations, it was only when thresholds used to identify cartilage damage were specific to severely involved knees (WORMS ≥4 or ≥5) that the prevalence of lateral PF damage consistently surpassed that of medial damage in knees with valgus malalignment.
To determine the extent to which involvement of the patellar crista accounted for the high prevalence of medial damage identified using WORMS, we reread 26 knee MRIs previously identified as having medial patellar disease with WORMS ≥2. In 23 of these knees (88.5%; 95% CI 76.2% to 100%), regions of the medial patella other than the crista had cartilage damage (see online supplementary figure S1), indicating that crista involvement alone was not sufficient to account for the high prevalence of medial PF disease in these cohorts.
Contrary to the Law of Valgus, our findings indicate that medial PF cartilage damage is highly prevalent in these older adult populations, and may even be more prevalent than cartilage damage in the lateral PF compartment. Only when identification of cartilage damage was limited to severely involved knees (WORMS ≥4 or ≥5) did the prevalence of lateral PF damage approximate that of medial damage. These findings were consistent across three large OA studies and regardless of whether neutral, varus or valgus aligned knees were considered. Scoring the crista as part of the medial patella appears to have had minimal influence on these results.
Non-mechanical influences may have more to do with the prevalence of medial PF disease than dominant theories admit. Yet, to the extent that mechanical factors do influence the distribution of PF cartilage damage, a review of PF biomechanics is warranted. The patella is a sesamoid bone embedded within the quadriceps tendon. Forces acting on the patella include the proximal pull of the quadriceps and the distal pull of the patellar tendon. Where these vectors deviate from 180°, we are left with a Q-angle (figure 1). In a normally aligned limb, the Q-angle is valgus, which results in greater lateral than medial PF loading. According to Ficat and Hungerford, it follows that medial PF disease should occur only rarely and in the presence of varus knee malalignment.
By outfitting knees with retropatellar pressure sensors under varying Q-angle conditions, cadaveric studies22 ,23 have shown that increased valgus Q-angles produce increased lateral PF loading. As suggested by others,24 varus knee malalignment has potential to invert the Q-angle from valgus to varus, bringing about greater medial PF loading and a higher prevalence of medial PF cartilage damage. However, our findings show that even among knees with neutral or valgus alignment, the prevalence of medial PF damage remained high and often exceeded that of lateral damage. Other factors must therefore be considered.
Recent studies using dynamic MRI provide detailed assessments of PF kinematics.25,–,28 In knee extension, the patella is superior and slightly lateral of the trochlear groove. As the knee begins to flex, the patella descends into the groove by first moving medially in relation to the femur.29 Once in the groove, patellar alignment is largely stabilised. However, at the initiation of flexion, patellar alignment is still susceptible to changes in tibial and/or femoral position.30,–,32Factors causing medial patellar displacement or tilt during the initiatory phase of knee flexion could contribute to medial PF cartilage damage as the patella enters the groove. One relevant influence may be foot pronation which, as hypothesised by others,33,–,36 can constrain the tibia to rotate internally37 ,38 and may thereby displace the patellar tendon medially, perhaps increasing medial PF load. Other potential influences include external femoral rotation or retroversion. The former might be actuated by toe-out walking.39
Regardless of whether an explanation is apparent, these findings may have implications for biomechanical treatments for PF OA. Several surgical procedures attempt to unload the symptomatic PF joint by realigning the patella medially. Commonly prescribed physical therapy exercises pursue similar objectives by preferentially strengthening the medial quadriceps. PF braces and taping procedures attempt to pull the patella medially and away from the supposed site of OA in the lateral PF compartment. None of these strategies are likely to have success if disease is localised to the medial rather than the lateral PF compartment, or if it affects both the medial and lateral compartments together. While specific advice must await the results of investigations focused on painful knees, the findings of the present study nevertheless caution against unconsidered prescription of treatments that assume PF OA to be exclusively or predominantly lateral.
A limitation of this study is that the PF joint was read using WORMS, for which the medial patella includes the crista. To evaluate the extent to which this aspect of WORMS influenced our findings, we had knees with medial patellar cartilage damage reread. The great majority (88.5%; 95% CI 76.2% to 100%) had cartilage damage in regions of the medial patella other than the crista. Extrapolated to the larger study population, crista damage alone might explain no more than 24% (upper CI) of the observed prevalence of medial patellar damage, still leaving a very large proportion (76–100%) unexplained by conventional theories.
Differences between individual OA study populations complicate attempts to generalise these findings to any single target population. A majority of subjects contributing to the current investigation were selected for study because they had or were at risk for knee OA. Only FOA subjects were a sample of the general population. While findings from the FOA cohort were consistent with findings from all other cohorts, only a subset of FOA participants contributed knee alignment measures. We suggest that the best outline of a target population for any particular study's findings is found in a careful review of that study's inclusion criteria.
We found that the prevalence of lateral PF damage approximated that of medial damage only for thresholds indicating severe cartilage damage (WORMS ≥4 or ≥5). In a longitudinal study of ROA, Cahue et al40 found that medial PF OA progression was less common than lateral PF OA progression. This may indicate that lateral disease, when present, is driven to progress more readily than medial PF disease. Based on the current cross-sectional study, we cannot determine the relative rates of progression in medial and lateral PF cartilage damage.
In conclusion, a rethinking of the aetiology of PF OA needs to account for the high prevalence of medial PF cartilage damage in these older adult populations.
The authors are indebted to participants and staff in the Boston Osteoarthritis of the Knee Study, the Framingham Osteoarthritis Study and the Multicenter Osteoarthritis Study.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Web Only Data - This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
- Web Only Data - This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Funding Provided by the National Institutes of Health.
Competing interests Dr Guermazi has received consultancy fees (more than $10 000 each) from Stryker and Novartis, and (less than $10 000 each) from Merck Serono, Facet Solutions and Genzyme. He owns stock and/or holds stock options in Synarc and is president of Boston Imaging Core Lab, LLC. Dr Guermazi was compensated for MRI readings that he performed as part of this study. Dr Roemer owns stock and/or holds stock options in Boston Imaging Core Lab, LLC. Dr Roemer was compensated for MRI readings that he performed as part of this study.
Patient consent Obtained.
Ethics approval This study was conducted with the approval of the Boston University Medical Center Institutional Review Board.
Provenance and peer review Not commissioned; externally peer reviewed
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.